Diabetic retinopathy (DR) is the leading cause of blindness in working-age Americans. There is evidence for a genetic contribution to DR~ the heritability of DR has been estimated to be as high as 27%. To date, only a few genes that might contribute to DR have been discovered. This is in part because genetic studies of DR have thus far contained limited numbers of participants. New methods of combining and analyzing samples from different racial/ethnic groups provide an opportunity to increase our ability to detect genes associated with DR. Hypotheses: Combining genetic samples from participants from multiple populations will increase the ability to discover DR genes. Use of novel admixture association, covariate modeling and fine-mapping methods will further maximize the power to detect associated genes. Specific Objectives: 1. To identify genetic variants associated to DR by performing a genome-wide association study (GWAS) using existing data from African American (AA) and Caucasian diabetic subjects and by incorporating admixture association into the GWAS. 2. To increase the power of genetic discovery by incorporating information about the participants' duration of diabetes and glycemic control into the GWAS using liability threshold modeling. 3. To replicate and fine-map the most promising associations using existing genetic data from DR participants from additional racial/ethnic groups. Methods: Existing genome-wide genotyping data from multiple AA and Caucasian cohorts of individuals with type 2 diabetes (T2D) and DR Phenotyping will be jointly analyzed using new statistical methods to combine single nucleotide polymorphism association and admixture association signals for increased power. Individuals with T2D and no DR with be compared to individuals with T2D and DR defined using the Early Treatment Diabetic Retinopathy Study scale. Subsequently, data on the two major non-genetic DR covariates, duration of diabetes and glycemic control, will be incorporated into the combined AA and Caucasian GWAS via liability threshold modeling. Finally, a set of genome-wide significant or suggestive variants (P-value < 10-5) from the GWAS analyses will be replicated in cohorts from additional populations, including South Asians and East Asians. All signals that attain genome-wide significance after replication analyses will be carried forward to fine-mapping. Fine-mapping will be performed without any additional genotyping using cross-population meta-analysis of genotyped and imputed HapMap 3 and 1000 Genomes variants, applying new metrics and methods developed for cross-population fine-mapping that optimally leverage the different patterns of linkage disequilibrium in different populations. Implications: If genes associated with DR are identified, they will provide insights into pathophysiology and may lead to new therapies for DR. In addition, results may provide information with which to counsel diabetic patients regarding their risk of DR.